CHDKPTP - PC Remote Control Performance Analysis - page 8 - RAW Shooting and Processing - CHDK Forum

CHDKPTP - PC Remote Control Performance Analysis

  • 465 Replies
  • 136504 Views
*

Offline reyalp

  • ******
  • 14126
Re: CHDKPTP - PC Remote Control Performance Analysis
« Reply #70 on: 02 / September / 2012, 21:07:58 »
Advertisements
In words: For me to shoot a scene at a desired calibrated ISO value I would say {shoot -sv=calibrated_sv}, which then CHDK converts for Canon to shoot at its own internal value that complies with the calibrated standard
I wouldn't assume the "real" ISO is calibrated. To determine that, you'd need to compare against some external standard.

I did try comparing my a540 auto exposure to my Sekonic L-38 hand held light meter, and it appears to agree more closely with the "real" ISO than the "market" values.
Quote
I gave -sv=.. 
     desired
 calibrated_sv        Internal ISO that is used by Canon firmware (the value that appears in EXIF)
         47                                    80  (the lowest VGA gain possible in an S90)
The value you refer to as "internal" is the "market" value. CHDK doesn't work with this value at all.

Quote
I am allowed an -sv= parameter range of 47 to 1944 without exceeding the gain bounds of the VGA circuit.  Correct?
Unknown, but in general the ISO override has no effect beyond the supported Canon range (once you account for the real vs market distinction), so it's probably OK.
Don't forget what the H stands for.

*

Offline SticK

  • *****
  • 779
Re: CHDKPTP - PC Remote Control Performance Analysis
« Reply #71 on: 02 / September / 2012, 21:58:20 »
Quote "I wouldn't assume the "real" ISO is calibrated. "

So then ... what would be the purpose of doing all the mapping (ie the table), unless there is some standardization objective at the end of the road? 

Quote "The value you refer to as "internal" is the "market" value. "

Yes ... because it has no absolute relationship with the "real" value ... it can be different, probably from manufacturer to manufacturer.

Quote "CHDK doesn't work with this value at all."

That's where I am confused ... where does the conversion by 1.6 take place then?  In Canon firmware?  For example when I say {shoot -sv=100} where does the 164 value come from?  My rational brain tells me that you would convert it and give Canon the value of 164.  Some other mechanism?

Another way of looking at this: assume I set the camera to the Canon value of 100 (on the LCD display).  If I press SHOOT button in CHDKPTP (or shoot on the camera), the shot really should be designated in EXIF as 63 (likely it's not possible but would be nice), not 100.  So to get the same intensity of exposure between SHOOT button and shoot command, I have to say {shoot -sv=63}.  I am sure you'll agree.  But because I was seeing both flavors of ISO which are mutually incompatible in my EXIFs at first (I was mixing SHOOT and shoot), that threw my head in a real knot.

The foregoing leads me to this:  let us say that "I don't care" about the "real" ISO value.  In other words, the values Canon uses I want to treat as "my personal standard values too," for consistency rather than real vs market when shooting from Canon or from CHDK {shoot}.  Therefore, I could make a function {myshoot -mysv=} that will execute the inverse relationship (ie set sv = mysv/1.6) at its front end (more accurately, using the data from your spreadsheet), in effect, asking CHDK to have a "market interface."  Then SHOOT and shoot would be consistent, but understandably, not "real." 

Is there a way I can do this easily without coding?  An option somewhere?






*

Offline reyalp

  • ******
  • 14126
Re: CHDKPTP - PC Remote Control Performance Analysis
« Reply #72 on: 02 / September / 2012, 22:58:39 »
So then ... what would be the purpose of doing all the mapping (ie the table), unless there is some standardization objective at the end of the road? 
It shows you the relationship between the values the Canon UI displays, and the values the firmware, CHDK and chdkptp work with. I strongly suspect that the "real" values have a consistent, direct relationship to the APEX definition across all the cameras, but I haven't proven it.

If you want some absolute, calibrated value, then calibrate it. You are developing imaging equipment, do you not have a calibrated light meter or light source?

It's probably safe to assume the Tv and Av are accurate, "real" units to reasonable precision (e.g. if you set Tv=1/1000th, you might get 1/1024th, but you won't get something like 1/600th). If you know the Bv from some external source (like a light meter), then you can easily derive the Sv from the APEX equation, and from there you can work out how it relates to the Canon ISO values. (Note that you can get a Bv value from the camera, but if the Sv differs from APEX standard, then the Bv must also)
Quote
That's where I am confused ... where does the conversion by 1.6 take place then?  In Canon firmware?
Yes! It's the Canon marketing display value, Canon firmware generates it. If CHDK knew the exact conversion (and knew it was correct across all the cameras), then we could offer overrides in the same units as the Canon OSD. We don't, therefore we work directly in the "real" values that Canon uses internally. We didn't pick "real" values to make your life difficult, we use them because those are the ones we know how to use!
Quote
For example when I say {shoot -sv=100} where does the 164 value come from?  My rational brain tells me that you would convert it and give Canon the value of 164. 
No. If you say shoot -sv=100, the value I send to the camera is 96*log2(100/3.125), which comes from the APEX definition. CHDK and chdkptp don't generate the "market" value, Canon does. As I said before, if CHDK knew how to use the "market" values, then I would just let you specify the market value.
Quote
The foregoing leads me to this:  let us say that "I don't care" about the "real" ISO value.  In other words, the values Canon uses I want to treat as "my personal standard values too," for consistency rather than real vs market when shooting from Canon or from CHDK {shoot}.  Therefore, I could make a function {myshoot -mysv=} that will execute the inverse relationship (ie set sv = mysv/1.6) at its front end (more accurately, using the data from your spreadsheet),
You should not assume the table I posted applies to any camera other than the SD990.
Quote
Is there a way I can do this easily without coding?  An option somewhere?
You could print out the table and tape it next to your console I guess ;)

It would be trivial to change shoot to send ISO/1.6 instead of ISO, but given that the relationship isn't well defined across all the  cameras and the 1.6 factor isn't valid at the lowest values, I would not make this a standard feature. I guess I could give you an "ISO conversion factor" set option or something.
« Last Edit: 02 / September / 2012, 23:01:10 by reyalp »
Don't forget what the H stands for.

*

Offline SticK

  • *****
  • 779
Re: CHDKPTP - PC Remote Control Performance Analysis
« Reply #73 on: 02 / September / 2012, 23:57:21 »
The concepts are beginning to grow on me. 

Quote "I would not make this a standard feature. I guess I could give you an "ISO conversion factor" set option or something."

Although I greatly appreciate your offer, but here is a place where I'd rather not request a feature that would not serve others as well.  I imagine there are some photographers who only own Canon digital cameras and publish the settings of their photos in the market Canon values.  I think they could find this useful too.  If you feel that's the case, then sure.

Quote "You are developing imaging equipment, do you not have a calibrated light meter or light source?"

You bet  ... but it's for my spectrometers not imaging.  I am not doing any calibration of the imaging component at present but it is possible I will do so in the future.   Because the imaging absolute value is not critical is why I asked about the conversion set feature.  Once the CCD is out and mounted, camera Av is out of the equation anyway (although an exposed CCD may have an intrinsic Av, it's not important).  If I were to calibrate the CCD, the calibration would take the whole instrument into account and in my case the Av would translate into a single flux value in W/m^2 of the specimen surface.  So for me, Canon ISO values are just easy to work with.

Quote "You should not assume the table I posted applies to any camera other than the SD990."

Yes I know now ... your comment relates to what I just said about "manufacturer to manufacturer."  I would use the k value from the empirically derived table I showed you already. 

Quote "the 1.6 factor isn't valid at the lowest values"

It would be easy to derive the table as I did, including a series of ks for the lowest value(s), and then do an interpolation in the non-linear portion at the low end.  What do you think? .. a kind of compromise?

Quote "As I said before, if CHDK knew how to use the "market" values, then I would just let you specify the market value"

Therefore it would be up to the individual user and his camera to generate a table like I did (my mission succeeded) and plug it into the front end of {shoot} (the set option) so that he could use {shoot} with ISO values he is familiar with, right?

*

Offline srsa_4c

  • ******
  • 4451
Re: CHDKPTP - PC Remote Control Performance Analysis
« Reply #74 on: 03 / September / 2012, 20:32:56 »

*

Offline SticK

  • *****
  • 779
Re: CHDKPTP - PC Remote Control Performance Analysis
« Reply #75 on: 03 / September / 2012, 21:12:44 »
The values in the 2nd column were tested, that is, entered into the -sv= in the shoot command.  The sought values are in the EXIF ISO column, and the attempt was to find sv by iteration such that it yielded the standard set of Canon internal ISO values (left column) in the DNG file.

for EXIF ISO:        set sv= to:
     80                      49
   100                      61
   200                    121.5
   400                    243
   800                    486
 1600                    972
 3200                  1944  (S90 limit)

k = 1.6461  for the S90.

Hence a single multiplier can be used in the S90 because translation is linear throughout the whole range, and a "camera sensitive" fudge factor has been established.

The EXIF ISO in the DNG file is sensitive to the exact value of sv=.  EXIF ISO in the JPG is more lenient, that is, any value close to one in the sv= column will produce a round number.  Thus I would say the EXIF ISOs in the DNG are monotonic, whereas EXIF ISOs in the JGPs are discrete.

Begs the question ....  what is the real exposure for a given shot if the DNG EXIF ISO shows up as 205 and JPG EXIF ISO shows up as 200?  Is it 205?  If monotonic, naturally there would have to be some dependence on VGA control DAC (resolution steps).

*

Offline SticK

  • *****
  • 779
Re: CHDKPTP - PC Remote Control Performance Analysis
« Reply #76 on: 03 / September / 2012, 21:26:49 »
@srsa_4c
Quote " "No image" issue followup:"
http://chdk.setepontos.com/index.php?topic=6321.msg65413#msg65413

Yes that makes sense and I can bring guard down so I can proceed at ease, thanks for this.  At present on the S90 the DNGs are in 100CANON and the JPGs are in 117___9.  So far so good since the reformat, the camera & CHDKPTP are stable, although I have not yet fully exercised everything I need to.  If by mistake I do set the combined files mode, I will run into same problem again.  There is a patch waterwingz refers to ... is that something I can implement on the S90, if not too complicated?


Re: CHDKPTP - PC Remote Control Performance Analysis
« Reply #77 on: 03 / September / 2012, 21:33:17 »
There is a patch waterwingz refers to ... is that something I can implement on the S90, if not too complicated?
AFAIK the only patch I mentioned was a way to change the DNG "bad pixel" level ?  So just to be clear,  that's unrelated to the other issues you mention in your post ?
Ported :   A1200    SD940   G10    Powershot N    G16

*

Offline reyalp

  • ******
  • 14126
Re: CHDKPTP - PC Remote Control Performance Analysis
« Reply #78 on: 03 / September / 2012, 21:37:23 »
S90 doesn't have CAM_DATE_FOLDER_NAMING implemented, it probably should.
Don't forget what the H stands for.

*

Offline SticK

  • *****
  • 779
Re: CHDKPTP - PC Remote Control Performance Analysis
« Reply #79 on: 03 / September / 2012, 21:48:58 »
Some useful CHDKPTP GUI stuff  ........

1) While direct transfer mode is not implemented yet (I know you are thinking about it), to simplify interaction, could you modify CHDKPTP to automatically update the Files tab after a shot is taken, while keeping my current directories open?  If the added file is out of view, scroll to put in view?  At present, I have to invoke Refresh and that closes my directories.

2) Could you enable mouse-pull up/down of the window pane separator bar?

3) Reminder -- after pressing SHUTDOWN, the DISCONNECT button should read CONNECT.  Otherwise behavior is correct despite the wrong label.

4) Is there a way to select a group of files within a directory for download?

5) edit: When I check/uncheck Viewfinder 1:1, the window does not go back to its original size by itself ... remains oversized.

=============

Quote "S90 doesn't have CAM_DATE_FOLDER_NAMING implemented, it probably should."

If not a big deal, it would prevent a mistake and having to reformat.
« Last Edit: 03 / September / 2012, 21:54:08 by SticK »

 

Related Topics


SimplePortal © 2008-2014, SimplePortal